Semantic Retrieval of Electronic Medical Records Based on Concept Networks

Abstract

In recent years, information and network technology has been commonly used in medical applications. Patient records are easily translated into electronic medical information by computer tools, thus providing more accurate, clearer, and more convenient retrieval of information. An electronic medical record often includes many medical concepts, which may represent important items in the record. If these concepts can be used to help information retrieval, semantic discrepancy can be reduced and retrieval precision can be greatly raised. In this paper, we thus propose an approach to construct concept networks from electronic medical records and to convert the semantic relations in the concept networks to semantic strengths. We also propose a modified evaluation approach with semantic consideration to find the match degree of two keywords. By the semantic relations of concept networks and the proposed evaluation approach, the match degrees between inquiry keywords and index keywords and between inquiry keywords and electronic medical records can be found. The medical records with higher matching degrees will then be output to users. Considering semantic relations and using the ontology can be expected to help improve the effectiveness and the flexibility of information retrieval.